Current model · ElevenLabs
Scribe v3 benchmarks, pricing and specs
Scribe v3 is an ElevenLabs model released on November 10, 2026 that takes audio input and produces text. It costs $0.003333 per audio minute. It scores 1.95 % WER on Artificial Analysis AA-WER v2, #1 of 8 current models (Artificial Analysis, measured November 15, 2026).
- Per audio minute
- $0.003333Listed today
- Artificial Analysis AA-WER v2
- 1.95 % WER#1 of 8 current models
Data verified November 15, 2026 · Dataset 2026-11-15
Specs
Key facts
- Provider
- ElevenLabs
- Model family
- Scribe
- Release date
- November 10, 2026
- Status
- Current
- Access
- API only
- API model id
- scribe_v3
- Context window
- Not reported
- Max output
- Not reported
- Input
- audio
- Output
- text
- Task families
- audio
- Declared capabilities
- speech to text, diarization, audio understanding
Charts
Against every current audio model
Where Scribe v3 lands among the current models of its task, on every independent benchmark, on price and on context.
Where it stands
On Artificial Analysis AA-WER v2, Scribe v3 ranks #1 of 8 current audio models. 2 of 16 list a lower price per audio minute. Each bar is the share of current models it beats.
Independent benchmarks
Artificial Analysis AA-WER v2
#1of 8
1.95% WER · ahead of 100%
Price and context
Price per audio minute
#3of 16
$0.003333 · cheaper than 80%
Benchmarks
Independent benchmark results
Scores come from independent evaluators that ran the model themselves, pinned to a dated snapshot. Rank is among current models in the same snapshot. ElevenLabs's own launch figures are listed under strengths with the publisher named, because they are not measured the same way.
| Benchmark | Score | Rank | Measured | Source |
|---|---|---|---|---|
| Artificial Analysis AA-WER v2 | 1.95 % WER | #1 of 8 | November 15, 2026 | Artificial Analysis · 2026-11-15 |
Fit
Strengths, cautions and best uses
- Strengths
- Artificial Analysis measures a 1.95% word error rate on AA-WER, the lowest of any transcription model in this catalogue, ahead of MAI-Transcribe-2 (2.04%) and Scribe v2 (2.18%)
- ElevenLabs reports 27% fewer word errors than Scribe v2 across 102 languages and the lowest speaker-diarization error of the models it tested, with 17% fewer errors than v2 on recordings with three or more speakers
- $0.20 per hour of audio, the same as Scribe v2; keyterm prompting takes up to 1,000 terms, and numbers can come back as digits or words
- Cautions
- Keyterm prompting and transcript editing cost extra: $0.05 and $0.07 per hour of audio
- ElevenLabs' comparisons use its own internal version of FLEURS and a meeting and call dataset it has not published yet
- Built for recorded audio: live transcription is the separate Scribe v3 Live model, at $0.22 per hour
- Best for
- Accurate batch transcription of meetings, calls and archives in many languages
- Transcripts that need reliable speaker labels
From benchmark to production
Building on Scribe v3?
Public scores rarely hold on your own data. We test it against the alternatives on your real tasks, then build the agent or pipeline around the model that wins.
AI Model Cost AuditLaunch price $100, regular price $150
Alternatives
Alternatives to Scribe v3
The closest current audio models by price. The cheapest model that scores as well or better is listed first; failing that, the cheapest one within 3% of the benchmark's scale.
On score and price
Scribe v3 beside the 4 alternatives listed below, on Artificial Analysis AA-WER v2 and on price per audio minute. One of them costs less and scores as well or better.
Artificial Analysis AA-WER v2
Measured November 15, 2026, higher is better
- Scribe v3
- 2
- Grok Voice Transcribe 2.0
- 2.3
- Voxtral Small Transcribe 3
- 2.1
- GPT Transcribe
- 3.3
- Gemini 3.5 Transcribe
- 2.6
Price per audio minute
Listed today, lower is better
- Scribe v3
- $0.00333
- Grok Voice Transcribe 2.0
- $0.00167
- Voxtral Small Transcribe 3
- $0.004
- GPT Transcribe
- $0.0045
- Gemini 3.5 Transcribe
- $0.005
Cheaper, similar score
Grok Voice Transcribe 2.0
- 50% cheaper per audio minute, $0.001667 vs $0.003333
- 0.3 points lower on Artificial Analysis AA-WER v2, 2.29 vs 1.95
- From xAI
Voxtral Small Transcribe 3
- 20% more per audio minute, $0.004 vs $0.003333
- 0.1 points lower on Artificial Analysis AA-WER v2, 2.09 vs 1.95
- From Mistral AI
GPT Transcribe
- 35% more per audio minute, $0.0045 vs $0.003333
- 1.4 points lower on Artificial Analysis AA-WER v2, 3.31 vs 1.95
- From OpenAI
Gemini 3.5 Transcribe
- 50% more per audio minute, $0.005 vs $0.003333
- 0.7 points lower on Artificial Analysis AA-WER v2, 2.6 vs 1.95
- From Google
FAQ
Scribe v3 FAQ
How much does Scribe v3 cost?
Scribe v3 costs $0.003333 per audio minute at the standard tier. It is the #3 cheapest of 16 current audio models priced per audio minute. Source: ElevenLabs, verified November 15, 2026.
How does Scribe v3 score on independent benchmarks?
Scribe v3 scores 1.95 % WER on Artificial Analysis AA-WER v2, #1 of 8 current models (Artificial Analysis, measured November 15, 2026). These are independent evaluations, not ElevenLabs's own launch figures.
Is Scribe v3 open source or API only?
Scribe v3 is available through ElevenLabs's API only; its weights are not published.
What are the best alternatives to Scribe v3?
The closest current alternatives by price and measured performance are Grok Voice Transcribe 2.0 (50% cheaper per audio minute, $0.001667 vs $0.003333), Voxtral Small Transcribe 3 (20% more per audio minute, $0.004 vs $0.003333), GPT Transcribe (35% more per audio minute, $0.0045 vs $0.003333) and Gemini 3.5 Transcribe (50% more per audio minute, $0.005 vs $0.003333).
AI Model Brief
The AI models worth knowing, once a week
What shipped, how it scores on independent benchmarks, what it costs per million tokens and what was deprecated, from the data behind our AI model benchmarks.
We use your address only to send the brief. See our privacy policy.
See this week’s briefSources
Sources
- Introducing Scribe v3 and v3 Live · ElevenLabs · official · verified November 15, 2026
- ElevenLabs API pricing · ElevenLabs · official · verified November 15, 2026
- Artificial Analysis speech-to-text leaderboard (AA-WER v2, non-streaming) · Artificial Analysis · independent · verified November 15, 2026 · Artificial Analysis Terms of Use
All 283 sources behind the benchmarks, with their versions, licenses and notes, are in the source catalog.
Cite this page
Van Data Team. "Scribe v3 Benchmarks, Pricing and Specs." AI Model Benchmark Dataset 2026-11-15. https://vandatateam.com/ai-model-benchmarks/scribe-v3
Reuse the data: JSON · CSV · or explore every model in the AI model benchmark comparison.
Production evaluation
Deciding whether Scribe v3 fits your workload?
Public benchmarks narrow the shortlist. We test the finalists on your own tasks, data and cost limits, then build the routing and review controls to run the winner in production.
- Evaluation cases and acceptance thresholds tied to real work
- Model routing, fallbacks and cost controls